Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

POWER MANAGEMENT STRATEGY FOR A PARALLEL

HYBRID ELECTRIC TRUCK
Chan-Chiao Lin* Huei Peng* J.W. Grizzle†
*Dept. of Mechanical Engineering, University of Michigan, MI 48109-2125, hpeng@umich.edu
†
Dept. of Electrical Engineering and Computer Science, University of Michigan, MI 48109-2122

Keywords: Hybrid Electric Vehicle, Power Management in terms of NOx (Nitrogen Oxides) and PM
(Particulate Matters) emissions. Recently, hybrid
Abstract electric vehicle (HEV) technology was proposed as
Hybrid vehicle techniques are widely studied the basis for new vehicle configurations. Owing to
recently because of their potential to significantly their dual on-board power sources and possibility of
improve the fuel economy and drivability of future regenerative braking, HEVs offer unprecedented
ground vehicles. Due to the dual-power-source potential for higher fuel economy while meeting
nature of these vehicles, control strategies based on tightened emissions standard, particularly when a
engineering intuition frequently fail to fully explore parallel configuration is employed. To fully realize
the potential of these advanced vehicles. In this the potential of hybrid powertrains, the power
paper, we will present a procedure for the design of management function of these vehicles must be
a near-optimum power management strategy. The carefully designed. The “power management”
design procedure starts by defining a cost function, function refers to the design of the higher-level,
such as minimizing fuel consumption and selected low-bandwidth control algorithm that determines
emission species. The Dynamic Programming (DP) the proper power level to be generated, and its split
techniques are then utilized to find the optimal between the two power sources. In general, the
control actions. Through analysis of the behavior of control sampling time for the power management
the DP control actions, sub-optimal rules are control system is low (~1Hz). Its command then
extracted, which, unlike DP control signals, are becomes the set- points for the servo-loop control
implementable. The performance of the power systems, which operate at a much higher frequency
management control strategy is verified by using (>20Hz). The servo-loop control systems can be
the hybrid vehicle model HE-VESIM developed at designed for different goals, such as improved
the Automotive Research Center of the University drivability, while ensuring the set-points
of Michigan. A trade-off study between fuel commanded by the main loop controller are
economy and emissions was performed. It was achieved reliably.
found that significant emission reduction can be
achieved at the expense of small increase in fuel Power management strategies for parallel HEVs
consumption. can be roughly classified into three categories. The
first type employs intelligent control techniques
1 Introduction such as control rules/fuzzy logic/neural network for
Medium and heavy trucks running on diesel estimation and control algorithm development ([1],
engines serve an important role in modern societies. [2]). The second approach is based on static
More than 80% of the freight transported in the US optimization methods. Commonly, electric power
in 1999 was carried by medium and heavy trucks. is translated into an equivalent amount of (steady-
The increasing reliance on the trucking state) fuel rate in order to calculate the overall fuel
transportation brings with it certain negative cost ([3], [4]). The optimization scheme then
impact. First, the petroleum consumption used in figures out proper split between the two energy
the transportation sector was one of the leading sources using steady-state efficiency maps.
contributors for import oil gap. Furthermore, diesel- Because of the simple point-wise optimization
engine vehicles are known to be more polluting nature, it is possible to extend such optimization
than gasoline-engine vehicles, schemes to solve the simultaneous
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

fuel economy and emission optimization problem
[5]. The basic idea of the third type of HEV control
algorithms consider the dynamic nature of the
system when performing the optimization ([6],[7]).
Furthermore, the optimization is with respect to a
time horizon, rather than for an instance in time. In
general, a power split algorithm resulting from
dynamic optimization approaches are more
accurate under transient conditions, but are more
computation-intensive.
In this paper, we apply the Dynamic Programming
(DP) technique to solve the optimal power
management problem of a hybrid electric truck. The
optimal power management solution over a driving
cycle is obtained by minimizing a defined cost
function. Two cases are solved: a fuel- economy
engine was downsized from V8 (7.3L) to V6 (5.5L)
and then augmented by a 49 KW DC electric motor.
An 18Amp-hour advanced valve- regulated lead-
acid (VRLA) battery was chosen as the energy
storage system. The hybrid truck was estimated to
be 246 kg heavier than the original design. A
schematic of the vehicle is given in Figure 1. The
downsized engine is connected to the torque
converter (TC), which in turn connects to the
transmission (Trns). The transmission and the
electric motor are linked to the propeller shaft (PS),
differential (D) and two driveshafts (DS). Important
parameters of this vehicle are given in Table 1.
Engine
Gas Drivetrain

cooler
only case, and a fuel/emission case. The

Inter
Air

comparison of these two cases provides insight into

the change needed when the additional objective of Control

emission reduction is included. However, the DP

control actions are not implementable due to their
Figure 1: Schematic diagram of the hybrid electric truck
preview nature and heavy computational
requirement. They are, on the other hand, DI Diesel Engine V6, 5.475L, 157HP/2400rpm
benchmarks other control strategies can compare DC Motor 49kW
against and learn from. We study the behavior of Capacity: 18Ah
Module number: 25
the dynamic programming solution carefully, and Lead-acid Battery
Energy density: 34 (Wh/kg)
extract implementable rules. These rules are used to Power density: 350 (W/kg)
improve a simple, intuition-based algorithm. It was Automatic Transmission 4 speed, GR: 3.45/2.24/1.41/1.0
found that the performance of the intuition (rule) Vehicle Curb weight: 7504 kg
based algorithm improves significantly, and in Table 1 Basic vehicle parameters
many cases approaches the DP optimal results. Load Input Data

The paper is organized as follows: In Section 2, the Load Output Variables

DIESEL ENGINE

hybrid electric truck model is described, followed cyc_m ph

DRIVELINE

by the explanation of the preliminary rule-based Dring Cycle

control strategy. The dynamic optimization DRIVER

HEV
Controller
BATTERY
ELECTRIC MOTOR

problem and the DP procedure are introduced in VEHICLE DYNAMICS

Section 3. The optimal results for the fuel

consumption and fuel/emissions optimization cases
Figure 2: Vehicle model in SIMULINK
are given in Section 4. Section
5 described the design of improved rule-based The Hybrid Engine-Vehicle SIMulation (HE-
strategies. Finally, conclusions are presented in VESIM) model used in this paper is based on the
Section 6. conventional vehicle model VESIM developed at
the University of Michigan [8]. VESIM was
2 Simulation Model (HE-VESIM) validated against measurements for a Class VI truck
2.1 System Configuration for both engine operation and vehicle
The baseline vehicle studied is the International launch/driving performance. The major changes
4700 series, a 4X2 Class VI truck. The diesel from VESIM include the reduction of the engine
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

size/power and corresponding fuel/emission map,

until the SOC reaches the upper limit, and then
and the integration of the electric components. The
Power Split rule will take over. The basic logic of
HE-VESIM model is implemented in SIMULINK, each control rule is described below.
as presented in Figure 2. Since the model has been Power Split Control: Based on the engine
efficiency map (Figure 3), an “engine on” power line,
presented before ([8], [9]), details are omitted here.
Pe _ on , and “motor assist” power line, Pm_ a , are
2.2 Preliminary Rule Based Control Strategy
chosen to avoid engine operation in inefficient
Many existing HEV power management areas. If Preq is less than Pe _ on , the electric motor

algorithms are rule-based, because of the ease in will supply the requested power alone. Beyond
handling switching operating modes. For parallel Pe _ on , the engine becomes the sole power source.

hybrid vehicles, there are five possible operating Once Preq exceeds Pm _ a , engine power is set at Pm _ a

modes: motor only, engine only, power-assist

and the motor is activated to make up the
(engine plus motor), recharging (engine charges
difference ( P - P ).
req m_a
the battery) and regenerative braking. Using
0
motor to start the engine occurs within short period 500
500 2
7 ..23 054 2.
.
26
11 6
0
.
2 002 4
0 2
of time and thus is not treated as a regular operating 450
450 .26
0.0
.21.
0
.2. 1
Power
Power
3 assist
400 2
2 4
00
mode. In order to improve fuel economy and/or to 400
350
12
2.
0
.
22
Pm_ a assist
Engine Torque (Nm)

0
reduce emissions, the power management ) 350
m 300 27 6
.2.02
0 4
5
3
0.2
(N 0. 1 6 6
controller has to decide which operating mode to que
300
250 0.22 .2
0 1
.2 4 .2
r 0.216 0
.2 0 32 0
use, and if proper, to determine the optimal split o
n
200
T 250
e 0.216
1 4 .
0
5
gi 2 0.2
. 2
between the two power sources while meeting the n
E
200
150 0.22
0.24
0
6
0.25
0.26 0.24 0.2
100 .23
driver demand and maintaining battery state of 150
0.24 0.23 0.27 Motor
0 .24 0 0.20.25
6 0.27
100
50
0.25
0.26 only
0.27 Motor
0.27
Pe_ on
charge. The simple rule-based power management 50 only
800 1000 1200 1400 1600 1800 2000 2200 2400
strategy to be presented below 800 1000 1200 1400 1600 1800 2000 2200 2400
Engine Speed (rpm)
Engine Speed (rpm)

was developed on the basis of engineering intuition
and simple analysis of component efficiency
tables/charts [9, 10], a very popular design
approach. The design process starts by interpreting
the driver pedal motion as a power request, Preq . The
operation of the controller is
determined by three simple rules: Braking rule,
Power Split rule and Recharging rule. If Preq is
Figure 3: Power Split Control rule
Recharging Control: In addition to power the
vehicle, the engine sometimes needs to provide
additional power to charge the battery. Commonly
a pre-selected recharge power level, Pch , is added
to the driver’s power request which
becomes the total requested engine power. The
motor power command becomes negative

negative, The Braking rule is applied to decelerate ( Pm  Pch ). However, this simple rule is frequently

the vehicle. If Preq is positive, either Power Split or found to be inefficient, and exceptions must be

Recharging rule will be applied, depending on the allowed. One example is that when Preq is less

battery state of charge (SOC). A high-level than Pe _ on , the recharging mode might not be

charge-sustaining strategy tries to maintain the
battery SOC within defined lower and upper
bounds. A 55-60% SOC range is chosen for
efficient battery operation as well as to prevent
battery depletion or damage. It is important to note
that these SOC levels are not hard bounds and
excursions could occur. Under normal propulsive
activated. If SOC is not excessively low, the motor
will still propel the vehicle to prevent inefficient
engine operation. The other exception is that when
Preq is greater than Pm_ a , the motor
power will become positive to assist the engine, or
stay at zero (when SOC is too low).
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.
Braking Control: When Preq is negative,
driving condition, the Power Split rule determines

the power flow in the hybrid powertrain. regenerative braking is activated. If Preq exceeds

Whenever the SOC drops below the lower limit, the regenerative braking capacity Pm_ min , friction
the controller will switch to the Recharging rule brakes will assist the deceleration ( P  P  P ).
b req m _ min
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.
2.3 Fuel Economy and Emissions Evaluation
Unlike light-duty hybrid vehicles, heavy-duty
hybrid vehicles do not yet have a standardized test
procedure for measuring their emissions and fuel
economy performance. A test protocol is under
development by SAE and NAVC based on SAE
J1711
3.1 Problem Formulation
[15]when we write this paper. Therefore, we In the discrete-time format, a model of the hybrid
decided to adopt the procedures proposed in [16]. electric vehicle can be expressed as:
The chassis-based driving schedule for heavy-duty
vehicles (UDDSHDV), as opposed to an engine-
only dynamometer cycle is adopted. For
UDDSHDV, emissions are recorded and reported
in the unit of gram per mile (g/mi). In addition, the
battery SOC correction procedure proposed in
[16] is used to correct fuel economy and emissions.
The hybrid electric truck with the preliminary rule-
based controller was tested through simulation over
the UDDSHDV cycle. Table 2 compares the results
of the HEV with those of the conventional diesel
engine truck. It can be seen that the hybrid-electric
truck, under the preliminary rule based control
algorithm, achieves 27% better fuel economy
compared to the baseline
diesel truck. PM reduction is also achieved even
though no emission criterion is explicitly included,
due to the trickling-down effect of improved fuel
economy. NOx level increases because the engine
works harder. In fact, this is exactly the main point
of this paper: it is hard to include more than one
objective in simple rule-based control strategies,
which is commonly driven by intuition and trial-
and-error. The simple control strategy is not
optimal since it is usually component-based as
oppose to system-based. Usually we do not even
know how much
This motivates the room
use ofisDynamic
left for improvement.
Programming
as an analysis and design tool.
FE (mi/gal) NOx (g/mi) PM (g/mi)
Conventional Truck 10.343 5.3466 0.5080
Hybrid Truck
13.159 5.7395 0.4576
(Preliminary Rule-Base)
Table 2: Performance comparison: conventional vs. HEV
3. Dynamic Optimization Problem
driving cycle, the optimal operating strategy to
minimize fuel consumption, or combined fuel
consumption/emissions can be obtained. A
numerical-based Dynamic Programming (DP)
approach is adopted in this paper to solve this finite
horizon optimization problem.
x(k  1)  f (x(k),u(k )) (1)
where u(k) is the vector of control variables such as
fuel injection rate to the engine, desired output
torque from the motor, and gear shift command to
the transmission. x(k ) is the state vector of the
system. The sampling time for this main-loop
control problem is selected to be one second. The
optimization goal is to find the control input, u(k) ,
which minimizes a cost function, which consists of
the weighted sum of fuel consumption and
emissions for a given driving cycle. The cost
function to be minimized has the following form:
N 1 N 1
 )    NOx(k )    PM (k ) (2)
J  Lx(k),u(k)  fuel(k
 
k 0 k 0
where N is the duration of the driving cycle, and L
is the instantaneous cost including fuel use and
engine-out NOx and PM emissions. For a fuel- only
problem,     0 , and   0 ,   0 for a simultaneous
fuel/emission problem.
During the optimization, it is necessary to impose
certain inequality constraints to ensure safe/smooth
operation of the engine/battery/motor. The four (or
more precisely, eight) constraints we imposed are:
e _ min  e (k)  e _max
Te _ min e (k )   Te (k )  Te _  (k ) 
max e
(3)
Tm _ min  m (k), SOC(k )   Tm (k )  Tm _ max  m (k ), SOC(k ) 
SOCmin  SOC(k )  SOCmax
where e is the engine speed, Te is the engine
Contrary to rule-based algorithms, the dynamic
optimization approach relies on a dynamic model to
compute the best control strategy. For a given
 Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.
torque, Tm is the motor torque and SOC is the
battery state of charge. In addition, we also
impose two equality constraints for the
optimization problem, so that the vehicle
always meets the speed and load (torque)
demands of the driving cycle at each sampling
time.
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

The above problem formulation does not impose of the torque converter, Tx and Td are the output
any constraint on terminal SOC, the optimization torque of the transmission and differential,
algorithm tends to deplete the battery in order to respectively. Rx and x are gear ratio and efficiency
attain minimal fuel consumption. Hence, a terminal
of the transmission, which are functions of the gear
constraint on SOC needs to be imposed:
number, gx
J    L  x(k ), u(k)   G(x(N ))
N 1

k 0
= N 1 fuel(k )    NOx(k )    PM (k ) (SOC(N )  SOC )2
(4) 3.2.3 Transmission

  f The automatic transmission is modelled as a ratio
k 0

where SOC f is the desired SOC at the final time device with gear number as the sole state. The
control (‘shift’) to the transmission is constrained to
(which is usually equal to the initial SOC), and  is take on the values of –1, 0, and 1, representing
a positive weighting factor. downshift, sustain and up-shift, respectively. The
3.2 Model Simplification gear shift dynamics are then described by:

The detailed HE-VESIM model (24 states) is not gx (k 1)  gx (k )  shift(k) (8)
suitable for dynamic optimization because its high 3.2.4 Motor/Battery

computation demand. Due to the selection of the
sampling time (T=1sec), dynamics that are much
faster than 1Hz could be ignored. By analyzing the
dynamic modes, it was determined that only two
state variables need to be kept: the transmission
gear number and the battery SOC. The
simplifications of the five sub-systems: engine,
driveline, transmission, motor/battery and
The electric motor characteristics are based on the
efficiency data obtained from the Advisor program
[17] as shown in Figure 4. The efficiency of the
motor is a function of motor torque and motor
speed, m  f (Tm , m ) . However, due to the battery
power and motor torque limit, the final motor
torque becomes:

vehicle are described below. Tm   

 min Tm,req , Tm,dis ( m ), Tbat ,dis (SOC,  m )  Tm,req  0 (9)


3.2.1 Engine
 
max Tm,req , Tm,chg ( m ), Tbat ,chg (SOC,  m ) Tm,req < 0

The engine dynamics are ignored and the output where Tm,req is the requested motor torque, Tm,dis

torque generated is from a look-up table with two and T

m,chg are the maximum motor torque in the
independent variables: engine speed and fuel

injected per cylinder/cycle [8]. The feed-gas NOx motoring and charging modes, and Tbat ,dis and

and PM emissions are functions of engine torque Tbat ,chg are the torque bounds due to battery current

and engine speed and are obtained by scaling the limit in the discharging and charging modes.
emission maps from the Advisor program [17].
250

3.2.2 Driveline 200

150

The driveline components are fast and thus were

Motor Torque (Nm)

100

reduced to static models. 50

0
2

 e 
T , T  T ( )  T (5) -50
0.93

p 
K ( )  t r r p -100
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.
 r 
T  T  T T ( , g ) R (g )  (g ) (6) -150
0 100 200 300 400 500 600 700 800

x t x,l1 x,l 2 t x x x xx Motor Speed (rad/s)

Td  Tx  Rc  Tm  c  Td ,l ( x )   Rd  d (7) Figure 4: Efficiency map of the DC motor

where T and T are pump and turbine torques, K Of all the sub-systems, the battery is perhaps the
p t
least understood. The reason is that the battery
and Tr are the capacity factor and torque ratio of performance—voltage, current and efficiency as

the torque converter, r  t / e is the speed ratio manifested from a pure electric viewpoint, is the
outcome of thermally-dependent electrochemical
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

processes that are not quite complicated. Various damping, Fr and Fa are the rolling resistance
models have been developed in the literature. But if force and the aerodynamic drag force,
we ignore thermal-temperature and transient M  M  J / r 2 is the effective mass of the vehicle

r v r d

effects (due to internal capacitance), the battery

and J r is the equivalent moment of inertia of the
model reduces to a static equivalent circuit shown
in Figure 5. The only state variable left in the rotating components in the vehicle.
battery is the state of charge (SOC):
3.3 Dynamic Programming Method
V  V  4(R  R )  T  
2 (10) Based on Bellman’s principle of optimality, the

SOC(k  1)  SOC(k) 
2(Rint  Rt )  Qb DP solution for the cost function shown in Eq.(4)

where the internal resistance Rint and the open is [11]:

Step N  1 :
circuit voltage Voc are functions of the battery J * N 1(x(N 1))  min L(x(N  1),u(N  1))  G(x(N )) (12)

SOC, Qb is the maximum battery charge and Rt is u ( N 1)

Step k , for 0  k  N  1
the terminal resistance. The battery plays an
J *k (x(k))  min L(x(k ),u(k))  J *k 1(x(k  1))  (13)
important role in the overall performance of HEV u (k )  
because of its nonlinear, non-symmetric and The recursive equation is solved backwards to find
relatively low efficiency characteristics. Figure 6 the optimal control policy, subject to the inequality
shows the charging and discharging efficiency of constraints shown in Eq. (3) and the equality
the battery. It can be seen that discharging constraints imposed by the driving cycle. A
efficiency decreases at low SOC and charging standard way to solve the above stated DP problem
efficiency decreases at high SOC region. Overall, numerically is to use quantization and interpolation
the battery operates more efficiently at low power ([11], [12]). The state and control values are first
levels in both charging and discharging. quantized into finite grids. At each step of the
Rint (SOC ) Rt
optimization search, the function Jk (x(k )) is
evaluated only at the grid points
Voc (SOC )
of the state variables. If the next state, x(k  1) , does
not fall exactly on to a quantized value, then the
values of J *k 1(x(k  1)) in Eq.(13) as well as G(x(N ))

Figure 5 Static-circuit battery model in (12) are determined through interpolation.

40 30 Despite the use of a simplified model, and a
35
25 quantized search space, the long time horizon
30
makes the above algorithm computationally
Discharging Power (kW)

Charging Power (kW)

20

expensive. In this research, we adopted two “tricks”

25

20
15

15 to acceleration the optimization search. First, from

the velocity profile of the driving
10
10

5
0.4 0.45 0.5 0.55
Battery SOC
0.6 0.65 0.7
5
0.4 0.45 0.5 0.55
Battery SOC
0.6 0.65 0.7 cycle, the required wheel torque Twh,req is

Figure 6: Efficiency maps of the lead acid battery: determined by inversely solving Eq.(11). The wheel
discharging (left) and charging (right)
speed  wh,req can be computed by feeding
3.2.5 Vehicle
The vehicle is modelled as a point- the required wheel torque to the vehicle model in
mass: order to include the wheel dynamics and slip effect.
Combining this procedure with the defined
 B v (k )  v (k ) F  F  v (k )   (11)
v (k  1)  v (k )  1  T wh (k )  wh v
  v
state/input grid, the vehicle model can be replaced
M  r 
2
v v r v (k ) r a v

r

d d v
by a finite set of operating points parameterized by
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.
where vv is the vehicle speed, Twh is the net wheel Twh,req and  wh,req . The second trick adopted is to

torque from driveline and hydraulic brake, rd is construct pre-computed look-up tables for the new
the dynamic tire radius, Bwh is the viscous states and instantaneous cost as a function of
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

quantized states, control inputs, and operating operation between two power movers for achieving
points. Once these tables are built, they can be used the best fuel economy. Additional 6% fuel economy
to update Eq.(13) efficiently by the vector improvement was achieved by the DP algorithm
operations in MATLAB [12]. (Table 3) as compared with values shown in Table
2.
4. Dynamic Programming Results FE (mi/gal) Fuel (g/mi) NOx (g/mi) PM (g/mi)
The DP procedure described above produces an
  0,  0 13.705 234.71 5.627 0.446
optimal, time-varying, state-feedback control
Table 3: Summary of DP results
policy, i.e., u *(x(k), k ) . It should be noted that DP
creates a family of optimal paths for all possible 4.2 Fuel Economy and Emissions Optimization
initial conditions. Once the initial SOC is To study the trade-off between fuel economy and
specified, the optimal policy will find a way to emissions, the weighting factors are varied:
achieve the minimal weighted cost of fuel

consumption and emissions while bringing the  0,5,10, 20, 40

(14)
final SOC close to the desired terminal value  0,100, 200,400, 600,800

( SOC f ). The optimal control policy was applied to Selected optimization results are shown in Figures
the HE-VESIM model for final evaluation. In the 8 and 9. The case of     0 corresponds to the
following, two cases are presented: fuel optimal fuel economy scenario. Figure 8 shows the
economy only, and simultaneous fuel/emission trade-off in NOx emissions and fuel economy.
optimization. Increasing  leads to significant NOx reduction
while causing a small fuel economy increase.
Veh Spd (MPH)

40
UDDSHDV
Actual Increasing  results in reduced PM (Figure 9) but
20
0
higher NOx emissions and lower fuel economy

0.58
0 100 200 300 400 500 600 700 800 900 1000 (Figure 8). The trade-off between NOx and PM can
be seen from Figure 9 where larger  tends to
SOC

0.56

0.54
decrease PM emission but increase NOx emission.
0 100 200 300 400 500 600 700 800 900 1000 More importantly, significant reduction in NOx and
100 PM emissions can be achieved at the price of a
Eng Pwr (kw)

50 small increase in fuel consumption.

0
5.6
0 100 200 300 400 500 600 700 800 900 1000 =10 =0
40 =400 =0
Mot Pwr (kw)

20 5.2
=20
=600
NOx Emissions (g/mi)

0
=40
-20 =1000 =5
4.8
=0
0 100 200 300 400 500 600 700 800 900 1000
Time (sec) =10
4.4 =0

Figure 7 Simulation results for the fuel-economy-only case =[0, 5 10, 20, 40], =0
=40 =10, =[0, 100, 200, 400]
4
=0 =20, =[0, 200, 400, 600]
4.1 Fuel Economy optimization results =40, =[0, 400, 600, 800, 1000]

The weightings in Eq.(4) are chosen to be 3.6

12.8 12.9 13 13.1 13.2 13.3 13.4 13.5 13.6 13.7

  0,  0,  5 106 for this case. The UDDSHDV Fuel Economy (mi/gal)

driving cycle is again used. The initial and terminal represent the optimal
SOC were both selected to be 0.57. Simulation
results of the vehicle under the DP policy are shown
in Figure 7. There is a small difference (<2mph)
between the desired vehicle speed (UDDSHDV)
and the achieved vehicle speed, caused by model
mismatch and the long sampling time (1 sec). The
engine power and motor power trajectories
 Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.
Figure 8 Fuel economy versus engine-out NOx
emissions The case   40,  800 seems
to achieve
a good trade-off--NOx and PM are reduced by
17.3 % and
10.3% respectively at a 3.67% penalty on fuel
economy. Simulation results of this case are
shown in Figure 10. The SOC fluctuates in a
wider range compared to the fuel-only case
(Figure 6). It can be seen that for the fuel-only
case, almost all the negative motor power are
from
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

regenerative braking. In other words, the engine where  is a positive weighting factor. Figure 11
seldom recharges the battery. Therefore, all shows the optimal gear position trajectories from
electrical energy consumed comes from DP for different  values. It can be seen that larger
regenerative braking. This implies that it is not  results in less-frequent gear shifting. The value
efficient to use engine power to charge the battery.
of   1.5 is chosen. From the DP
This is due to the fact that the fuel efficiency map
of this diesel engine is flat in medium to high power results, the gear operational points are expressed on
regions. the engine power demand vs. transmission speed
0.49 plot (Figure 12). It can be seen that the gear
=[0, 5 10, 20, 40], =0

=40 =20
=10, =[0, 100, 200, 400]
=20, =[0, 200, 400, 600]
positions are separated into four regions and the
=0 =0 =40, =[0, 400, 600, 800, 1000]
0.46 boundary between adjacent regions represent
PM Emissions (g/mi)

=5
=0
optimal gear shifting thresholds. After adding a
=0
0.43 =0 hysteresis function to the shifting thresholds, a new
gear shift map is obtained. It should be mentioned
0.4
=10
that the optimal gear shift map can also be
constructed through static optimization ([10], [14]).
=1000
0.37
Given an engine power and wheel speed, the best
3.8 4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8
NOx Emissions (g/mi) gear position for minimum weighted cost of fuel
Figure 9 Engine-out PM emissions versus NOx emissions and emissions can be chosen based on the
0.62
0.6 combined steady-state engine fuel consumption and
0.58
SOC

0.56 emissions map. It is found that the steady- state gear

0.54
0.52 map from this method nearly coincides
100 200 300 400 500 600 700 800 900 1000
100 with Figure 12.
Eng Pwr (kW)

80
60 5
40 4  =0
20
3
Gear

0
100 200 300 400 500 600 700 800 900 1000 2
40 1
Mot Pwr (kW)

20 0
0 100 200 300 400 500 600 700 800 900 1000
0
5
-20
4 =0.5
-40 3
Gear

100 200 300 400 500 600 700 800 900 1000
Time (sec) 2
1

Figure 10 Simulation results (   40,  800 ) 0

0 100 200 300 400 500 600 700 800 900 1000
5. Development of Improved Rule-Based 5
4 =1.5
Controls 3
Gear

2
1
The DP control policy is not implementable in real 0
0 100 200 300 400 500 600 700 800 900 1000
driving conditions because it requires knowledge of Time (sec)

future speed and load profile. Nonetheless, Figure 11 Optimal gear position trajectory
analyzing its behaviour provides useful insight into 100
1st gear

possible improvement. 90 2nd gear

3rd gear
80 4th gear
Engine Power Demand (kW)

5.1 Gear Shift Control 70

60
Gear shifting strategy is crucial to the fuel economy 50
of hybrid electric vehicles. The original optimal 40

gear trajectory has frequent shifting which is 30

undesirable. Hence, a drivability constraint is added 20

to avoid frequent gear shifting: 10

N 1 0
0 20 40 60 80 100 120 140 160 180 200 220

J    fuel(k)  40  NOx(k)  800  PM (k)    gx (k  1)  gx (k)  (15) Transmission Speed (rad/s)

k 0

 5 106  (SOC(N )  SOCf )2 Figure 12 Gear operating points of DP optimization

Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

550
5.2 Power Split Control
500
In this section, we study how Power Split Control 450

of the preliminary rule-based strategy can be 400

Engine Torque (N-m)

Peng
PSR 
350
improved. A power-split-ratio is defined
Preq 300
250

to quantify the positive power flows in the 200

500

powertrain, where Peng is the engine power and Preq 150

is the power request from the driver. Four positive- 100

50
power operating modes are defined: motor-only (
PSR  0 ), engine-only ( PSR  1 ), power-assist ( 0 800 1000 1200 1400 1600 1800 2000 2200 2400 2600
Engine Speed (rpm)

 PSR  1), and recharging mode Figure 14 BSFEC map in g/kWhr

( PSR  1 ). The optimal (DP) behavior uses the

5.3 Charge-Sustaining Strategy
motor-only mode in low power-demand region The power split control scheme described above
when the vehicle launches. When the wheel speed does not maintain the battery SOC within desired
is above 6 rad/s, a simple rule is found by plotting operating range. The DP procedure is repeated
the optimal PSR versus the power request over the again except this time the regenerative braking
transmission input speed, which is equivalent to function is turned off. The curve-fitted optimal PSR
torque demand at the torque converter output shaft result is compared with the one with regenerative
(see Figure 13). The figure shows the optimal braking in Figure 15. More aggressive spending of
policy uses the recharging mode ( PSR  1 ) in the battery energy can be used when SOC is high and
low torque region, the engine-only mode in the more conservative rules can be used when SOC is
middle torque region, and the power-assist mode in low. These adaptive PSR rules can be learned from
the high torque region. This can be explained by DP results by specifying different initial SOC
examining a weighted Brake Specific Fuel and points.
Emissions Consumption (BSFEC) of the engine. 4

Wf   WNOx   WPM 3.5

BSFEC  (16) With Regenerative braking
Peng 3
Power Split Ratio (PSR)

Without Regenerative braking

The contour of engine BSFEC map is shown in the 2.5

Figure 14. It can be seen that the best BSFEC region 2

occurs at low torque level. In order to move the 1.5

engine operating points towards a better BSFEC 1

region, the engine is used to recharge the battery at 0.5

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

low load, and the motor is used to assist the engine Power Demand / Trans Speed (kN-m)

at high load. In order to extract an implementable Figure 15 Optimal PSR rules comparison
rule, a least-square curve fit is used to approximate
5.4 Performance Evaluation
the optimal PSR, shown as the solid line in Figure
After incorporating all the changes outlined in the
13.
4 previous sections, the improved rule-based
3.5 controller is evaluated using several different
3
driving cycles. In addition to the original cycle
Power Split Ratio (PSR)

2.5
(UDDSHDV), the new rule-based controller is put
2
through three other driving cycles (suburban,
interstate, and city) to test its robustness. The results
1.5
are shown in Tables 4-7. It can be seen that
1
depending on the nature of the driving cycles, the
0.5
new rule-based control system may not improve all
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Power Demand / Trans Speed (kN-m)

Figure 13 DP power split behavior (UDDSHDV cycle) three categories of performance, and in certain
cases did worse. However, if the
Proceedings of the 2002 Mediterranean Control Conference, Lisbon, Portugal, July 2002.

combined fuel/emission performance is considered References

(the “performance measure”), the new rule-based [1] Baumann, B. M. et al. “Mechatronic Design and Control of
Hybrid Electric Vehicles,” IEEE/ASME Transactions on
controller is always significantly better than the Mechatronics, v5 n 1 2000. p 58-72, 2000
original, intuition driven rule-based control law. [2] Farrall, S. D. and Jones, R. P., “Energy Management in an
Automotive Electric/Heat Engine Hybrid Powertrain Using
Performance
FE (mi/gal) NOx (g/mi PM (g/mi)
Measure * Fuzzy Decision Making,” Proceedings of the 1993 International
Baseline Rule-Based 13.159 5.7395 0.4576 840.63 Symposium on Intelligent Control, Chicago, IL, 1993
New Rule-Based 12.8738 4.8355 0.4292 787.0965 [3] Kim, C., NamGoong, E., and Lee, S., “Fuel Economy
DP (FE & Emis) 13.237 4.6422 0.3992 739.56 Optimization for Parallel Hybrid Vehicles with CVT,” SAE Paper
Table 4: Results over the UDDSHDV cycle No. 1999-01-1148.
[4] Paganelli, G. et al. “A General Formulation for the Instantaneous
Performance Measure: fuel  40  NOx  800  PM (g/mi)
Control of the Power Split in Charge-Sustaining Hybrid Electric
Vehicles.” Proceedings of 5th Int’l Symposium on Advanced
NOx PM Performance Vehicle Control, Ann Arbor, MI, 2000.
FE (mi/gal)
(g/mi) (g/mi) Measure
[5] Johnson, V.H., Wipke, K.B. and Rausen, D.J., “HEV Control
Baseline Rule-
Based
15.3103 4.4291 0.3547 671.225 Strategy for Real-Time Optimization of Fuel Economy and
New Rule-Based 14.5839 2.9273 0.2959 574.6322 Emissions,” Proceedings of the Future Car Congress, April 2000,
DP (FE & Emis) 15.4108 2.7785 0.2585 526.666 SAE Paper No. 2000-01-1543.
[6] Brahma, A., et al. “Dynamic Optimization of Mechanical
Table 5 Results over the WVUSUB cycle
Electrical Power Flow in Parallel Hybrid Electric Vehicles” Proc.
FE (mi/gal) NOx (g/mi) PM (g/mi)
Performance of 5th Int’l Sypm. Adv. Veh. Ctrl, Ann Arbor, MI, 2000.
Measure [7] Zoelch, U., and Scroeder, D., “Dynamic Optimization Method for
Baseline Rule-Based 12.8433 7.2850 0.5086 948.8256 Design and Rating of the Components of a Hybrid Vehicle,”
New Rule-Based 12.7198 6.2733 0.4878 894.106 International Journal of Vehicle Design, v19, n1, p1-13, 1998
DP (FE & Emis) 12.9658 6.1675 0.4411 847.6675
[8]Assanis, D.N. et al. “Validation and Use of SIMULINK
Table 6 Results over the WVUINTER cycle Integrated, High Fidelity, Engine-In-Vehicle Simulation of the
Performance International Class VI Truck,” SAE Paper No. 2000-01-0288
FE (mi/gal) NOx (g/mi) PM (g/mi) [9] Lin, C.C., Filipi, Z.S., Wang, Y., Louca, L.S., Peng, H., Assanis,
Measure
Baseline Rule-Based 16.1791 3.8698 0.3320 621.2218 D.N., and Stein, J.L., “Integrated, Feed-Forward Hybrid Electric
New Rule-Based 15.3649 2.4091 0.2187 480.7421 Vehicle Simulation in SIMULINK and its Use for Power
DP (FE & Emis) 16.6257 2.0367 0.1608 403.578 Management Studies”, SAE Paper No. 2001-01-1334
Table 7 Results over the WVUCITY cycle [10] Bowles, P. D., “Modeling and Energy Management for a Parallel
Hybrid Electric Vehicle (PHEV) with Continuously Variable
6. Conclusions Transmission (CVT),” MS thesis, University of Michigan, Ann
Arbor, MI, 1999
Designing the power management strategy for HEV [11] Bertsekas, D.P., Dynamic Programming and Optimal Control,
by learning from the Dynamic Programming (DP) Athena Scientific, 1995
results has the clear advantage of being near- [12] Jun-Mo Kang et al. “Approximate Dynamic Programming
Solutions for Lean Burn Engine Aftertreatment,” Proc. of the
optimal, accommodates multiple objectives, and IEEE Conf. on Dec. and Ctrl, Phoenix, AZ, Dec. 7-10, 1999
systematic. Depending on the overall objective, one [13] Lee, H.D. et al. “Advanced Gear Shifting and Clutching Strategy
can easily develop power management laws that for Parallel Hybrid Vehicle with Automated Manual
Transmission,” IEEE Ind. Appl. Conference, v3, 1998
emphasize fuel economy, and/or emissions. By [14] Soltic, P., and Guzzella, L., "Optimum SI Engine Based
analyzing the DP results, improved rule-based Powertrain Systems for Lightweight Passenger Cars," SAE
control strategy can be developed. The learned Paper No. 2000-01-0827, 2000
[15] Society of Automotive Engineers, Hybrid-Electric Vehicle test
behavior was found to be robust, rather than cycle- Procedure Task Force, “SAE J1711, Recommended Practice for
specific. This is evident by the fact that the learned Measuring Exhaust Emissions and Fuel Economy of Hybrid-
behavior based on one cycle works extremely well Electric Vehicles,” 1998
[16] Mckain, D. L. et al. "Characterization of Emissions from Hybrid-
for several never-seen driving cycles, moving the Electric and Conventional Transit Buses," SAE Paper 2000-01-
rule-based control law closer to the theoretical 2011, 2000
optimal (DP results) by 50-70%. [17] National Renewable Energy Laboratory, “ADVISOR 3.2
Documentation,” http://www.ctts.nrel.gov/analysis/ , 2001.
Acknowledgments
This research is supported by the U.S. Army
TARDEC under the contract DAAE07-98-C-R-
L008. The work of J.W. Grizzle was supported in
part by NSF contract IIS-9988695.